简体   繁体   English

R中的向量化迭代循环

[英]vectorizing iteration loop in R

Folks, I want to translate the following Visual Basic code into R: 伙计们,我想将以下Visual Basic代码转换为R:

'''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''
Function WetBulb(T As Double, WDes As Double, PAtm As Double)
' Function to calculate wet-bulb temperature from dry-bulb
' and humidity ratio
Dim Wsat As Double
Dim TWBOld As Double
Dim WOld As Double
Dim TWBNew As Double
Dim TWB As Double
Dim WStar As Double
Dim W As Double
Dim slope As Double
Wsat = HumRatRH(T, RHMax, PAtm)
TWBOld = T
WOld = Wsat
TWBNew = TWBOld - 1
Do
    TWB = TWBNew
    WStar = HumRatRH(TWB, RHMax, PAtm)
    W = ((HfgRef - (CpWat - CpVap) * TWB) * WStar - CpAir * (T - TWB)) / (HfgRef + CpVap * T - CpWat * TWB)
    slope = (W - WOld) / (TWB - TWBOld)
    TWBNew = TWB - (W - WDes) / slope
    If Abs(W - WDes) < Abs(WOld - WDes) Then
        WOld = W
        TWBOld = TWB
    End If
Loop Until Abs((TWBNew - TWB) / TWB) < tolRel
WetBulb = TWB
End Function

The difficulty that I have is the loop involves a vector, so I need to somehow vectorize this loop and also the if statement. 我遇到的困难是循环涉及一个向量,因此我需要以某种方式向量化该循环以及if语句。 Below is my attempt, but I think I only vectorize one of the two that I need to vectorize. 以下是我的尝试,但我认为我只对我需要向量化的两个向量之一进行了向量化。 I have included all the necessary functions and constants so that the snippet will run. 我已经包含了所有必需的函数和常量,以便片段可以运行。 The function is in the bottom. 该功能在底部。 I also have included a snippet test code with the correct answer. 我还提供了带有正确答案的代码段测试代码。

Any help is greately appreciated. 非常感谢任何帮助。

# Constants independent of unit system
NMol   = 0.62198       # ratio of molecular weights, Mvap/MAir
RHMax  = 1             # maximum relative humidity, 1 or 100 (if percent)
tolRel = 0.000001      # relative error tolerance for iteration

# Constants for English Units
# Note: constants currently configured for PAtm in atmospheres
HfgRef = 1061          # heat of vaporization at 0C, Btu/hr.lbm.F
CpVap = 0.444          # specific heat of water vapor, Btu/hr.lbm.F
CpWat = 1              # specific heat of liquid water, Btu/hr.lbm.F
CpAir = 0.24           # specific heat of dry air, Btu/hr.lbm.F
RAir = 0.02521         # gas constant for air, (user pressure).ft3/lbm.R
kPaMult = 101.325      # multiplier to get kPascals from user pressure
TAbs = 459.67          # add to user temperature to get absolute temp
TKelMult = 0.555556    # multiplier to get Kelvin from user temp
TAmb = 70              # typical temperature in user units (initial value)
#####################################################################
SatPress <- function(TArg) {

# Define constants for vapor pressure correlations
C1  = -5674.5359
C2  = -0.51523058
C3  = -0.009677843
C4  = 0.00000062215701
C5  = 2.0747825E-09
C6  = -9.484024E-13
C7  = 4.1635019
C8  = -5800.2206
C9  = -5.516256
C10 = -0.048640239
C11 = 0.000041764768
C12 = -0.000000014452093
C13 = 6.5459673

T = (TArg + TAbs) * TKelMult
# Use different correlations for pressure over ice or water
    kPa.lo = exp(C1 / T + C2 + T * C3 + T * T * (C4 + T * (C5 + C6 * T)) + C7 * log(T))
    kPa.hi = exp(C8 / T + C9 + T * (C10 + T * (C11 + T * C12)) + C13 * log(T))
kPa = ifelse(T < 273.15, kPa.lo, kPa.hi)
SatPress = kPa / kPaMult
return(SatPress)

}
#####################################################################

HumRatRH = function(T,RH,PAtm) {
# function to calculate humidity ratio from temperature
# and relative humidity
pw = SatPress(T) * RH / RHMax
HumRatRH = NMol * pw / (PAtm - pw)
return(HumRatRH)
}
#####################################################################
WetBulb = function(T, WDes,PAtm) {
# Function to calculate wet-bulb temperature from dry-bulb
# and humidity ratio
Wsat = HumRatRH(T, RHMax, PAtm)
TWBOld = T
WOld = Wsat
TWBNew = TWBOld - 1
iterate.TWB = function(x) {
    repeat {
    TWB = TWBNew
    WStar = HumRatRH(TWB, RHMax, PAtm)
    W = ((HfgRef - (CpWat - CpVap) * TWB) * WStar - CpAir * (T - TWB)) / (HfgRef + CpVap * T - CpWat * TWB)
    slope = (W - WOld) / (TWB - TWBOld)
    TWBNew = TWB - (W - x) / slope
    TWBOld=ifelse(abs(W - x) < abs(WOld - x),TWB,TWBOld) # update TWBOld first
    WOld=ifelse(abs(W - x) < abs(WOld - x),w,WOld)       # then update WOld
    if (abs((TWBNew - TWB) / TWB) < tolRel) break()
    }
    return(TWB)
}
WetBulb = sapply(WDes, iterate.TWB)
return(WetBulb)
}

#####################################################################

temp = c(80,55,100)
w = c(0.011,0.009,0.016)
PAtm = 0.8187308
WetBulb(temp,w,PAtm)

# The correct answer:
# 62.95381538   51.3986312   74.02877887

The easiest way to vectorise a function f is to use Vectorize . 向量f的最简单方法是使用Vectorize By default, it vectorises f with respect to all its arguments. 默认情况下,它会针对所有参数对f进行矢量化处理。 In this case, you only want to vectorise it for 2 out of 3 arguments, so you specify this via vectorize.args . 在这种情况下,您只想对3个参数中的2个进行矢量化,因此可以通过vectorize.args指定。

WetBulb <- Vectorize(WetBulb, vectorize.args=c("T", "WDes"))

(And you can also remove the sapply inside WetBulb .) This isn't necessarily the most efficient way to get vectorisation (it's basically syntactic sugar for a mapply call) but it's certainly the simplest. (并且您也可以删除sapply WetBulb 。)这不一定是进行矢量化的最有效方法(它基本上是mapply调用的语法糖),但肯定是最简单的方法。

声明:本站的技术帖子网页,遵循CC BY-SA 4.0协议,如果您需要转载,请注明本站网址或者原文地址。任何问题请咨询:yoyou2525@163.com.

 
粤ICP备18138465号  © 2020-2024 STACKOOM.COM